Measurement finished and reported in the 3 attached figures.

I finally choose to use this colormap for all plots as it is (I think) nice looking, color-blind and printing friendly.

Mean absorption seems to be 18+/-5 ppm/cm.

The sample was oriented to have the kanji for 'up' at the top.

Two HEPA capsules are installed for the air gun of PCI cleanroom. The maximum air pressure was specified to be 3.4 bar (figure 1). So I adjusted the pressure of the compressed air as attached figure2.

History: Marc found that there is air leakage between the connection part of capsule and tube. The connection was done by only fixing capsule and tube with steel belt. Aso-san put some white sealing tapes in this connection part. The use of sealing tape solves the problem of air leakage.

Yesterday I started the measurement of the tama size sample T1.

Thanks to the help of Simon and Matteo I could identify the good sample.

Indeed there are now 2 tama size sample in PCI clean room but one seems to be different (maybe FC mirror?)

The one I installed is the one on the left of entry 301.

However, it seems that this sample creates at least 2 beams in reflection that hit directly the last prism on the probe path before the sample.

In turn, multiple reflections are sent a bit everywhere (but not towards the photo detector).

Using 50 mW of incident pump power I could estimate the power lost in these reflections to be around 10% of the incident power.

Because the sizes of these reflections are quite large, we started the measurement anyway.

These measurements will finish this evening and I will turn off the pump laser at that time.

The 3 measurements (XY, XZ and YZ) have been finished and are reported on the attached figures.

Note that for the XZ and YZ measurements the mean and standard deviation are computed only inside the sample while the contour plot and histogram show the entire measurement.

The results is compatible with previous measurements and gives absorption ~ 60 +/- 5 ppm/cm without peculiar feature on XY plane.

I like a lot the plots in gray scale but I also did the same with colored scale.

I'll upload soon a notebook to do these plots.

The mirror is oriented to have a black dot on the top.

EDIT : changed number of bins in histogram

Yuhang, Michael

We measured the relative intensity noise of the new Mephisto 2W NdYAG laser to be used at TAMA. The laser was sent into a 75mm lens, then a 50:50 beamsplitter going to a photodetector and power meter sensor. The distance between the laser and detectors is fairly short due to the initial divergence of the laser beam. With the photodetector we took the following spectra, while the power meter was used to measure long term power stability. Some ND filters were placed between the laser and beam splitter to counteract saturation of the PD and prevent back reflection into the laser. 

First PD: Thorlabs PDA36A-EC 

Second PD: Qubic PD-AC100-Si 1-100 MHz

Figure 1 shows the low frequency laser RIN at approximately the minimum current required for the laser to start emitting (about 0.8 A), compared to the PD dark noise. The measurement was passed through the Stanford pre-amp at DC coupling and 100 gain.

Figure 2 shows the RIN at different laser current (power). The DC level on the detector changes on the highest two measurements due to the PD saturating. The gain on the pre-amp was reduced by a factor of half for the highest two current measurements. Due to the PD reading being close to the noise floor, this may be responsible for the jump between the levels of the measurements between different pre-amp gain values.

Figure 3 shows the effect of activating the noise eater. In our first measurement its effect only became noticeable at laser current > 1.55 A.

Figure 4 shows the predicted frequency dependence of RIN from the Coherent Mephisto manual. Ideally the noise eater suppresses the large peak and reduces the 10k-100kHz noise floor by ~15 dB. We don't see much if any of the latter effect though. It seems like the low frequency signal is too close to the detector noise level. The levels are a bit different from the measured results - I didn't apply normalization to the measurements from the spectrum analyzer readings, but the relative values for noise eater on/off are what is important.

To follow up, we tried changing the preamp gain settings and changing the spectrum analyser coupling from DC to AC to see if we could get better SNR. The Stanford pre-amp was close to saturating though.

Figure 5 shows the result of increasing the PD gain and optimising the pre-amp settings. We can get a bit more clearance from the PD dark noise versus figure 1, but still less than desireable.

We then tried to do another measurement using the second PD, but it seemed like it had more dark noise than the Thorlabs one just from inspection on the oscilloscope. The Qubic PD is AC only, but we couldn't see anything on the oscilloscope when applying 1-10 kHz modulation, just ~ 5 mV rms noise.

Today I checked the bulk reference and got : R_bulk = 0.7527 cm/W (note that in this case I moved the IU by 0.32 mm)

With laser diode current of 6A I set up the HWP so to have P_in = 7.072W.

Using the DC values drop, the Kasi sample center positions are : X_center = 329.62 mm and Y_center = 122.81 mm.

The Zaber limits have been changed accordingly with some margin to check the incident power from remote.

I did a long Z scan and used the phase transition to get the center Z position at Z_center = 64.1 mm.

I realigned the DC at this position and started a circular map in the XY plane with parameters :

X = X_center, Y=Y_center, Z =Z_center, 20 mm radius, 0.1 mm step size, 500 ms waiting time, median/average filter order 5 and sensitivity = 5mV (ie 5 nA).

Preliminary absorption is : 75 ppm/cm (coherent with previous measurement)

with :

AC =4.6e-4 V
DC = 4.144 V
P_in = 7.074 W
P_t = 6.096 W
T_sample = P_t/P_in;
mat_correction=3.34;

All measurements should be finished tomorrow.

Katsuki, Marc

We performed several scans of the the z positions of the translation stage or the IU with the surface reference sample.

We plan to check if the shape of these signals could be used to perform more efficiently the alignment.

We tried to place back the PBS for the birefringence measurement but it was quite lossy (~20%).

Actually the one we used does not have name but the holder height seems to match. So we installed it as in entry 1496.

I'm planning to buy a new PBS as I could not find other.

The R_surf is now 16.79 /W at z_translationStage = 40.5 mm and z_IU = 66 mm.

Tomorrow we'll start the absorption measurement of the korean sample.

Michael and Yuhang

We received a new Mephisto laser this Monday. Michael and I tested lasing current threshold, noise eater engagement current threshold. We measured RIN with different frequency band, current, noise eater on/off, and power stability. In this entry, we report new Mephisto power stability.

We used a power meter to monitor the power stability of new Mephisto. The total monitoring time is 17.5 hours. The power change is attached in the figure. A peak to peak 1.4% power stability was found with irregular power change found at the end of measurement. This shows that laser power doesn't become really stable after long time operation.  

The IR detuning map in elog2615 was concerned that it could be related with alignment control. Therefore, I try to clarify this concern here and do some related test.

How is the AA loop?

     As far as I found, the AA loop bring GR and IR transmission to top of TEM00. If I introduce pertubation to AA loop, GR power goes down, which means AA loop works well.

Why IR detuning map should not be related to AA?

    1. The IR detuning map in elog2615 should not be related to AA loop because the AA loop works to have a totally linear response as elog2650. But IR detuning map has a flat region.

    2. I took IR detuning spectrum with AA loop opened. Meanwhile, the AA error signals are put around zero by hand. The beam hitting position was chosen as elog2573 to old and new spot. Then I got two spectrums are the attached figure. The spectrum shows result as found in elog2573. Since AA loop is open, there should be not effect from AA loop.

It is found, as attached figure 2, that PR and BS mirror had a sudden position change on last Thursday (20210812). This makes the filter cavity alignment totally lost. Since the movement of PR mirror is so large that the PR mirror picomotor needed to be used to recover alignment.

The mirror movement has coincidence with an earthquake (figure 1). But it is the first time I notice that mainly PR pitch mirror is moved. BS is moved a bit as well. But no obvious mirror movement is found for input and end.

The tire of the bicycle in the south arm was broken. However, the one in the west arm was OK. So I exchanged the place of them. In the future, we can use a good bicycle (blue color as attached picture) to go to the end room of south arm.

I measured CCFC error signal for 3 hours with 30 minutes interval (Fig. 1). Fig. 2 shows the CCFC error signals around 0 crossing point. The 0 crossing point of the fitting result (dashed curves) changed by (1.04-0.97)*54 = 3.8 Hz in 3 hours.

I also measured the nonlinear gain and shot noise before/after the whole measurement.

The nonlinear gain changed from 4.7 to 4.3 in 3 hours. This nonlinear gain change causes the CCFC amplitude change. The normalized CCFC amplitude can be written as follows.

Normalized CCFC amplitude = x/(1-x^2)^2 = (1-1/sqrt(g))/(2/sqrt(g)-1/g)^2

Figure 3 shows the normalized CCFC amplitude as a function of nonlinear gain. When the nonlinear gain changes from 4.7 to 4.3, the normalized CCFC amplitude changes from 1.07 to 0.97 by a factor of 0.9. In fact, the CCFC amplitude changed from 132mVpp to 118mVpp by a factor of 0.9 in 3 hours.

I measured CCFC error signal with different length LEMO cables for CCFC LO (attached figure). The CCFC amplitude was 132mVpp. The red and green curves in the figure represent the CCFC error signal with red and brown+green LEMO cables for CCFC LO, respectively. As you can see, the CCFC error signal with brown+green LEMO cable is close to I phase. So I will use the brown+green cable for CCFC LO.

I measured CCFC FDS with fixed homodyne angle for 3 hours with 30 minutes interval (figure 1). The FC was unlocked between each FDS measurement. According to the least square fitting, the detuning changed by 7 Hz in 3 hours even with fixed homodyne angle.

Before each FDS measurement, I optimized p pol PLL frequency to have maximum BAB transmission with 20mW green. The nonlinear gain change was 4.4-4.7 in 3 hours, which corresponds to generated squeezing of 10.1-10.5 dB. Since I optimized p pol PLL every time I measured the nonlinear gain, this nonlinear gain change is the real nonlinear gain change, not the detuning change of BAB.

I also measured shot noise before/after all the FDS measurement (figure 2). The shot noise changed by 0.15dB in 3 hours.

sqz_dB = 10.5;                   % generated squeezing (dB)

L_rt = 120e-6;                   % FC losses

L = 0.49;                        % propagation losses  

A0 = 0.06;                       % Squeezer/filter cavity mode mismatch

C0 = 0.02;                       % Squeezer/local oscillator mode mismatch

ERR_L =   1.5e-12;               % Lock accuracy (m)

ERR_csi = 30e-3;                 % Phase noise (rad)

Marc, Matteo

During the measurement of ETMY HR surface absorption we had to make large Z motions to move ETMY around safely.

At one point, the Z position got stuck at the away sensor position on Z.

We found that this was due to the fact that the 2 translation stages along Z have a little shift between them. This makes the 2 away sensors position slightly different on the 2 translation stages.

There was therefore a little twist between the 2 motors and the lockstep safety prevented any further motions (even when relaxing the lockstep twist tolerance).

The solution was :

Remove the lockstep

Move the motor individually so that they have the same position

Put the lockstep back

Bring the 2 motors to the home position (this can be performed if the vertical position allows to go above the laser bench)

Katsuki-san, Marc

We placed the ETMY jigs, jacks and metal/teflon rings inside a plastic box.

We used goro-goro to clean the floor and optical bench.

We moved several samples on the small table below the small clean booth.

We tidy up many cables and removed a malfunctioning usb hub. Now we don't have any more connections troubles with polarizers or powermeter.

We reinstalled the imaging unit (z_IU=68mm) and started to check the calibration factor with the surface reference sample.

We found that the maximum is with the translation stage at Z = 40.2 mm : R_surf = 16.6 /W.

We took several measurements between Z = 38.5 mm to 41.5 mm. The goal of these measurements is to find this Z position but also to be used later on to check if the lateral peaks could be useful to perform quicker alignment (to be compare with OSCAR simulations).

The next step is to scan the IU position.

We also brought the SHINKOSHA evaluation plate #7 to the storage room.

[Aritomi, Yuhang, Michael]

We measured CCFC FDS with fixed homodyne angle for 3 hours with 30 minutes interval (figure 1). We fixed OPO temperature to 7.163kOhm and p pol PLL frequency to 200MHz.

According to the least square fitting, the detuning changed by 9Hz in 3 hours even with fixed homodyne angle.

We measured shot noise before/after all the FDS measurement (figure 2). The shot noise is the same in 3 hours.

We also measured nonlinear gain after each FDS measurement. The nonlinear gain changed from 4.6 to 4.2 in 3 hours, which corresponds to generated squeezing of 10.3 dB to 9.8 dB. 

Regarding the nonlinear gain measurement, we divided the BAB transmission with 20mW green by that without green. For BAB transmission with 20mW green, we fixed p pol PLL frequency to 200MHz and locked OPO and measured the maximum value of BAB transmission by scanning CC1 with 20mW green. The reason why we fixed the p pol PLL frequency is that it was fixed during the FDS measurement. For BAB transmission without green, we scanned OPO and measured the peak value of BAB transmission.

There are two mechanisms which change the BAB maximum with 20mW green. One is the nonlinear gain change and another is optimal p pol PLL frequency change (in other words, s&p overlap inside OPO or BAB detuning inside OPO). In the nonlinear gain measurement above, s&p do not always overlap inside OPO with 20mW green (BAB can be detuned inside OPO), while they overlap without green. This means that in this method, the measured nonlinear gain change is due to both of the real nonlinear gain change and optimal p pol PLL frequency change.

To measure the real nonlinear gain change, we need to optimize p pol PLL frequency every time we measure the nonlinear gain to make sure s&p overlap with 20mW green.

sqz_dB = 10;                      % generated squeezing (dB)

L_rt = 120e-6;                    % FC losses

L = 0.52;                         % propagation losses   

A0 = 0.06;                        % Squeezer/filter cavity mode mismatch

C0 = 0.02;                        % Squeezer/local oscillator mode mismatch

ERR_L =   1.5e-12;                % Lock accuracy (m)

ERR_csi = 30e-3;                  % Phase noise (rad)

I am very sorry that I sent a wrong information to Aritomi-san.

The optical losses information Aritomi-san listed is actually detuning in Hz. (If you compare this 'wrong' losses with detuning in legend of mcmc figure, you can see they are the same)

The real optical losses are attached in this comment. They change from 49% to 54%.

The detuning measurement had some problems. The main issue is that pointing loop was not kept to be zero while these scanning.

When pointing loop is kept to be zero, either we introduce pertubation to input or end mirror yaw/pitch, the beam hitting position can change only on input mirror. Therefore, when we do this scan, we are doing a map similar to elog2615. But instead of end mirror in elog 2615, this scan is for input mirror. Attached figure 1 shows the schematic of this scanning process.

Since the PR/BS pointing loop has angular scanning range of about 200urad, which can scan a range only about 0.8mm on input mirror. After doing these scan on input/end mirrors y/p, we got detuning change, input/end mirror oplev signal, and pointing error signal as attached figure 2-5.

With position change of about 0.8mm on input mirror, the detuning change is less than 5Hz. (This is a detuning change smaller than the flat region found on filter cavity end mirror in elog2615.)

Matteo and Yuhang

The filter cavity detuning was found to be changing when we lock filter cavity with GR (elog2642, elog2640, elog2636). The reason is that the correction signal sent to main laser sensed by GR is not exactly IR required.

However, we found out if we change main laser frequency manually by large amount (500MHz), the detuning almost doesn't change before and after main laser frequency change. This is already observed in elog2636. In elog2636, the first unlock didn't make correction signal change, which tells us that main laser frequency was changed by ~40MHz and the re-lock is fast enough that cavity length didn't change. Therefore, in the re-lock, we saw the detuning changes continously. However, the second unlock of elog2636 had correction signal going back almost to zero. We can see that around this un-lock, there were some oscillations, which makes the re-lock have time to cross ~40 FSR and arrives at the position where correction signal can be close to zero. In the first un-lock, cavity is locked to the same FSR, but different FSR is locked during the second un-lock, which makes the detuning go back to zero. This is because the same frequency change is required for GR and IR when we go to other resonances.

Instead of sending correction signal below 0.1Hz to end mirror, we can send it to main laser temperature. This makes the acquire of resonance easier and faster. The stability of lock is also more stable in this case, comparing to send all signal to main laser. In addition, we know the filter cavity length should change less than few tens of micro-meters per day. But the main laser frequency is expected to have drift more than 100MHz per day (from PLL observation). Even from the laser manual, the stability of laser frequency DC value is described to be changed by less than 1MHz per minute with a constant room temperature. So it seems to be reasonable to take filter cavity length as a reference below 0.1Hz.

With this locking strategy, I monitored filter cavity detuning stability by about 20hours on 23rd July. A screen shot of this monitor is attached in this elog without calibration. Upper: correction signal to main laser temperature. Lower: detuning. In the first 15 hours, the detuning stability is about 5Hz. But similar to elog2636, there is detuning change happend in the last five hours, which seems to come from some other detuning change mechanisms. One example is that the AOM which is currently used will have frequency drift of 220Hz after one year.

Notice: AA and pointing loops are always closed.